Skip to content

Metamorphic Testing for LLM #98

@AmileshN

Description

@AmileshN

Purpose

Ensure CityAgent produces consistent results under semantic-preserving input transformations.

Tools

  • pytest
  • Parameterized tests(come up with this)

Tasks

  • Implement metamorphic tests for:
    • Case-insensitivity
    • Order-independence
    • Acronyms (e.g., Dr vs Drive)
    • Synonyms
  • Compare structured outputs and citations, not exact wording

Definition of Done

  • Tests exist under tests/metamorphic/
  • Each invariant has at least one test case
  • Assertions validate semantic equivalence
  • Tests pass

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    Status

    Todo

    Milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions